Anthropic runs its own wet lab so its AI models can test biology ideas

By VK

Published 21 Sep 2026, 13:17 UTC

Reporting from TechCrunch

Anthropic confirmed it runs a Bay Area biology lab that lets its AI models carry out hands-on experiments. The company says the work centers on fundamental biology rather than drug discovery.

What it means for founders

Labs are becoming vertical players. A frontier lab with its own bench, a biotech acquisition and pharma partnerships is building capabilities that overlap with AI drug discovery startups, even if it says its focus is basic research.
Watch the customer boundary. Anthropic appears keen, per TechCrunch, not to look like a competitor to pharma clients. Startups building biology products on its models should track where that line moves.
Gated access is a new channel. The Life Sciences Verification Program suggests the most capable models for biology may come with vetting requirements, which affects timelines for teams planning to rely on them.
Biosecurity scrutiny will rise. With Amodei himself flagging bioterrorism risk, expect closer attention to any company pairing AI models with physical lab work.

The story

Anthropic is running its own wet lab for biology, located in the Bay Area, so that its AI models can be put to work on hands-on experiments, the company confirmed to TechCrunch. It would not say what the lab is working on. Anthropic says the main focus is fundamental biology, not drug discovery.

Why a model maker wants a lab

AI leaders have promised that the technology will help cure disease, and Anthropic chief executive Dario Amodei has said it could cure most major diseases within five to ten years. A language model can propose hypotheses, but checking them still means running experiments on real samples. A lab gives Anthropic a place to close that loop.

Eric Kauderer-Abrams, Anthropic's head of life sciences, told Reuters that real lab work remains the final test in biology. "We absolutely are doing that today," he said, according to the Reuters interview cited by TechCrunch. He described the setup as similar to most biotech labs, with Anthropic running some research itself and some alongside outside partners.

The move follows Anthropic's April acquisition of Coefficient Bio, an AI biotech startup that had been operating in stealth.

Keeping clear of its customers

TechCrunch suggests the emphasis on basic science has a commercial logic: pharmaceutical companies are among Anthropic's larger customers and partners, and it recently announced a joint drug discovery deal with Novo Nordisk. Anthropic has previously drawn criticism for launching products that customers saw as competing with their own.

The same week, Anthropic launched a Life Sciences Verification Program that gives vetted researchers access to its most capable models. It has also published reports on work supporting drug discovery, including one on speeding up protein design and one on biomolecular modeling.

An awkward backdrop

The lab arrives during a tense stretch for the company. Researcher Jacob Coxon resigned with a public warning about how seriously people building AI take its dangers, and Anthropic's alignment lead has put the chance of AI wiping out humanity within the next decade at above 10%. Over the weekend before the lab news, Amodei wrote that AI companies should ease their pace and adopt rules for themselves, and he has repeatedly named bioterrorism as one of AI's biggest risks.

Some in tech have pointed out the tension between those warnings and a company running biology experiments. Investor Chamath Palihapitiya mocked the plan on X, saying he did not recommend it.